Time-dependent reliability analysis of package under non-Gaussian excitation

In the process of distribution, the package is often excited by non-Gaussian excitation. In this paper, a methodology was proposed for accurate simulation of non-Gaussian excitation as well as the efficient analysis of time-dependent reliability of packages. The non-Gaussian excitation was simulated based on the first four moments of the input, i.e. the mean, variance, skewness, kurtosis and the PSD (or autocorrelation) using the polynomial chaos expansion and the Karhunen-Loeve expansion. In order to analyze the package reliability in an efficient way, the quasi Monte Carlo method was adopted, in which, the random variables were reasonably selected in random variable space to simulate non-Gaussian excitation, the acceleration response of package was obtained using the 4th Runge-Kutta method. The accurate statistical information of the first four moments and the autocorrelation can be obtained by reduced number of simulation. Based on the moments and autocorrelation of the acceleration response, the proposed polynomial chaos expansion, the Karhunen-Loeve expansion and the quasi Monte Carlo method were used to simulate the package response. the time dependent reliability of package was analyzed and the analysis was verified by crude Monte Carlo simulation. The proposed non-Gaussian vibration simulation method and package reliability analysis method can provide theoretical basis for package analysis and packaging design optimization. © 2020, Editorial Office of Journal of Vibration and Shock. All right reserved.